We investigate how historically accumulated entomological knowledge can be made machine-accessible, semantically interlinked, and reusable while preserving its provenance and context.
Entomological literature, archival sources, and collection records preserve information on species, localities, and traits, as well as on people and scientific networks. Much of this information is available only in unstructured texts, historical terminology, or images.
Drawing on entomological sources, we develop and evaluate methods for making historically accumulated bodies of knowledge accessible and interlinking them. Our central research questions are: How can taxonomic information be reliably extracted despite historical fonts and typefaces, OCR errors, multilingual content, and changing nomenclature? How can people, species, places, publications, and collection objects be uniquely identified and linked while preserving their provenance and historical context? And how should digital knowledge infrastructures be designed to ensure that the resulting data remain reusable in the long term?
To this end, we combine text mining, natural language processing, AI-assisted text recognition and image analysis with knowledge graphs, persistent identifiers, and standards such as RDF and Darwin Core. These methods are tested on entomological sources and evaluated using taxonomic expertise. Human oversight remains essential when results are uncertain, as does the responsible handling of sensitive knowledge resources and those shaped by colonial contexts.
As a unit of the Senckenberg Data and Modelling Center, we develop transferable methods for entomological collections and knowledge infrastructures.
Historical sources broaden the temporal and contextual scope of digital collections and offer new avenues for investigating long-term changes in insect diversity.
